Your team’s way of working, carried to every AI tool.

Correct one AI, and every AI your team uses follows it. The corrections you already make become your team’s decisions: no playbook to write, nothing new to learn.

Early access is open to teams and individuals, and every account is set up personally.

Works withClaudeChatGPTMicrosoft 365 CopilotGeminiClaude CodeCodexCursorAll integrations

Your AI doesn’t know what your team already decided

Most of how your team works was settled long ago and never written down. You notice when you hand a task to AI and have to explain yourself. Then a teammate explains it again, to their own assistant, in a different tool.

  • Said to Claude

    “No exclamation marks in anything a customer reads.”

    In effectSaved by Claude

    Don’t use exclamation marks in anything a customer reads.

    WHEN
    Writing anything a customer will read
    WHY
    It reads as unserious in support replies

    Followed by Claude, ChatGPT, Copilot and Gemini

  • Said to Copilot

    “Never promise a delivery date. Say we’ll confirm it within a day.”

    In effectSaved by Copilot

    Don’t promise delivery dates. Say we’ll confirm within one business day.

    WHEN
    Any reply about an order

    Followed by every assistant you use

  • Said to Gemini

    “Every proposal ends with next steps and a date.”

    In effectSaved by Gemini

    End every proposal with next steps and a date.

    WHEN
    Writing or editing a proposal

    Followed by Claude, ChatGPT, Copilot and Gemini

  • Said to ChatGPT

    “We say customers, not users.”

    Team rulebookPromoted by an admin

    Say customers, not users.

    WHEN
    Any writing about the people who buy from us
    FROM
    Saved by three members in ChatGPT

    Followed by every member’s assistants

How it works

  1. 01

    We set it up with you

    Request early access, and we connect it to the AI tools your team already uses, together with you. That’s the only setup.

  2. 02

    Correct them as you work

    Work as usual. When you tell an assistant how you want something done, it saves that as a decision, with when it applies and why.

  3. 03

    Every assistant follows it

    Before acting, each assistant checks the decisions that apply to the task. Correct Claude once, open ChatGPT, and it already knows.

Good practice spreads on its own

A shared rulebook

Team rules every member’s assistants follow, next to each person’s own decisions. Where the two disagree, the person’s own comes first.

Corrections become team rules

When several people keep making the same correction, admins see it and can make it a rule for everyone.

New people start ahead

A new hire’s assistant knows the team’s conventions from their first day.

Use cases by team

You stay in control

Everything your assistants follow is written down where you can see it, change it or switch it off. Here’s what we keep and who sees it.

Risky saves wait for you

If a save contains an email address, a link, a phone number or an instruction to send something somewhere, no assistant follows it until you confirm. It may have come from an email or web page your assistant read, not from you.

What we store

Your decisions, with when they apply, why, the words they came from, and which assistant saved them. Kept until you delete them.

Not your work

Assistants tell us only what task they’re about to do, so we can pick the right decisions. We don’t keep it, apart from a 10% sample for quality checks, deleted after 30 days. We never see your documents, emails or chats.

Model training

We don’t sell your data, show ads, or use your decisions to train AI models.

Who sees what

Admins can read what members save with a work account, and members are told before they join. Members never see each other’s decisions. Personal accounts stay personal.

See and undo everything

Every decision shows which assistant saved it, why, and in whose words. Retire any of them and bring it back later.

Export and delete

Export everything at any time. Deleting your account removes it, your sign-ins and your personal decisions for good.

Who processes your data
Neon and Fly.io store and run the service in the United States. TypeSafe, Voyage AI, OpenRouter and Anthropic read tasks and decisions to choose which apply. Google handles sign-in only and receives none of your decisions.

Trust & privacy

Changelog

  1. [Date]
    [What shipped][One line on what it changes for a team]
  2. [Date]
    [What shipped][One line on what it changes for a team]
  3. [Date]
    [What shipped][One line on what it changes for a team]

Ask your AI about us

Have your assistant read this site and tell you what Decision Cartography does for a team.

Read decisioncartography.com and explain in a few sentences what Decision Cartography does for a team.

Bring your team in early

We’re setting up a small number of teams and people one at a time, each one with you. Tell us what you keep explaining to AI.

Request early access